If $f$ is locally Lipschitz on $X$ and $X$ is compact, then $f$ is Lipschitz on $X$.

My proof: Since $f$ is locally Lipschitz on $X$, for each $x ∈ X$ there exists an open $B_x$ containing $x$ such that $f$ is Lipschitz on $B_x$. Consider the collection of all such $B_x$. This collection forms an open cover of $X$ and so there is a finite sub-collection $\{B_1, B_2, . . . , B_n\}$ which also covers $X$, since $X$ is compact. Since $f$ is Lipschitz on each $B_i$, there is an $B_i$ such that $d(f(x_i), f(y_i)) \leq M_i d(x_i, y_i)$ for all $x_i, y_i \in Wi$, for $i ∈ \{1, 2, . . . , n\}$. Then taking $B = max\{B_1, B_2, . . . , B_n\}$, we see that $f$ is Lipschitz on $X$.

Is my solution correct?

Can we use the fact of compact metric space that every sequence in $X$ has a convergent sub-sequence to prove the same fact??

  • 1
    $\begingroup$ I think there's a small problem with your proof. What if I pick $x,y$ belonging to two different balls $B_i$ and $B_j$? $\endgroup$ – TorsionSquid Feb 16 '15 at 2:56

First note that locally Lipschitz implies continuous. This said, the argument suggested can be completed as follows. For every $x\in X$ there is an open ball $B_x$ centered at $x$ of radius $\varepsilon_x$ and a constant $M_x$ such that $d(f(x),f(y))\le M_xd(x,y)$ for $x,y\in B_x$. Now consider the open convering consisting of the open balls $B'_x$ of center $x$ and radius $\varepsilon_x/2$, and by compactness extract a finite subcovering $B'_{x_1},\dots,B'_{x_r}$. In this situation: $$ \delta=\min\big\{\frac{\varepsilon_{x_i}}{2}: i=1,\dots,r\big\}>0\quad\text{and}\quad K=\max\{d(f(x),f(y)): x,y\in X\} $$ exists by compactness and continuity. Then pick a positive constant $$ L\ge \frac{K}{\delta}, M_{x_1},\dots,M_{x_r}. $$ We see that $f$ is $L$-Lipschitz. Indeed,

(1) if $d(x,y)<\delta$, as $x\in B'_{x_i}$ for some $i$ we have $$ d(y,x_i)\le d(y,x)+d(x,x_i)<\delta+\frac{\varepsilon_{x_i}}{2}\le\varepsilon_{x_i}, $$ hence $x,y$ are both in $B_i$ and $$ d(f(x),f(y))<M_id(x,y)\le L\,d(x,y), $$

(2) if $d(x,y)\ge\delta$ we have $$ d(f(x),f(y))\le K=\frac{K}{\delta}\delta\le L\,d(x,y). $$ We are done.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.